This article provides a brief overview of the latest research on the ability of forecasters to predict recessions. Few recessions have been forecast before their onset. Forecasters tend to be excessively cautious and do not revise their forecasts promptly and sufficiently to reflect incoming news. Nor do they fully take into account interdependence among economies. There is also a tendency for “groupthink” among forecasters, preventing them from giving due weight to their individual priors.

Abstract

This article provides a brief overview of the latest research on the ability of forecasters to predict recessions. Few recessions have been forecast before their onset. Forecasters tend to be excessively cautious and do not revise their forecasts promptly and sufficiently to reflect incoming news. Nor do they fully take into account interdependence among economies. There is also a tendency for “groupthink” among forecasters, preventing them from giving due weight to their individual priors.

Research Summaries

Most U.S. recessions remain undetected until they are well under way. This “predictive failure” was documented several decades ago by Zarnowitz (1986) and then by Fintzen and Stekler (1999). During the latest recession, which according to the National Bureau of Economic Research began in December 2007, the initial forecast for 2008 by private analysts in January 2007 was on average for growth of 3 percent. Almost every month since then the forecast has been lowered, but even as late as September 2008 forecasters expected on average that growth would be close to 2 percent. It was only in the last quarter of the year that expectations of growth adjusted sharply downward. Actual growth in 2008 was about 1 percent.

It is somewhat surprising that there is no evidence that the practical implications of the inefficiency of consensus forecasts are well understood by forecasters themselves

Data on private analysts’ forecasts of output growth are available for a broad range of advanced, emerging, and developing economies. Consensus Economics, Inc. has been collecting and publishing monthly forecasts by private analysts since October 1989 for major advanced economies under the title of Consensus Forecasts, and over time the data set was expanded to include many emerging and developing economies. Consensus Forecasts survey a number of prominent financial and economic analysts and report their forecasts as well as simple statistics summarizing the distribution of forecasts, particularly the mean (the “consensus”) and the standard deviation of the consensus (the “dispersion,” a measure of the extent of disagreement among forecasters).

Loungani (2001) used Consensus Forecasts to examine the track record for forecasting recessions in a diverse sample of advanced, emerging, and developing economies. He concluded that forecasters’ ability to predict recessions is generally very limited. Only two of the 60 recessions that occurred around the world during the 1990s were predicted a year in advance. Two-thirds of those recessions remained undetected seven months before they occurred. And even as late as two months before each recession began, about a quarter of the forecasts still did not predict a recession. Evidence from the 2000s examined by Loungani, Stekler, and Rodriguez (2008) suggests the recessions that occurred during this decade went also largely undetected until they started. (For the analysis of the track record for forecasting recoveries, see Loungani, 2002.)

Loungani, Stekler, and Tamirisa (forthcoming) explore forecasting performance for the recessions caused by economic and financial crises. They find that forecast errors for the recessions following banking crises exceed those for regular recessions, while the opposite is true for recessions following currency and debt crises. One reason for the greater predictive failure in the case of crisis-related recessions is a greater tendency of forecasters to smooth their forecasts, failing to adjust them sufficiently in response to news. The failure to incorporate foreign news, especially news from major emerging economies, appears to be because it is more of a challenge than incorporating domestic news. Forecasters do not take into account the dependence of economies on one another, particularly the closer linkages between advanced and emerging economies. These findings are broadly consistent with the results obtained for the G-7 economies by Isiklar, Lahiri, and Loungani (2006) in a paper in which the authors proposed a methodology for testing how quickly forecasters incorporate foreign news.

Another reason for the failure to predict recessions appears to be a tendency for herd behavior in forecasting, possibly owing to forecasters putting a higher weight on the group’s shared view than on private priors and incoming news. Such a tendency is particularly pronounced in forecasts of advanced economies, as reflected in the decline in the dispersion of consensus forecasts over the year preceding recessions. In contrast, for emerging and developing economies, the dispersion of consensus forecasts tends to rise about nine months before the start of a recession. This suggests that monitoring trends in the dispersion of forecasts may help improve forecasting performance for recessions.

Dovern, Fritsche, and Slacalek (2009) also find that disagreement about real variables (GDP, consumption, investment, and unemployment) intensifies strongly during recessions, including the current one. Disagreement over nominal variables (interest rates and inflation) rises with their level and is considerably lower under independent central banks. Cross-sectional dispersion for both groups increases with uncertainty about the underlying actual indicators, though to a lesser extent for nominal series. These findings suggest that more credible monetary policy can substantially contribute to anchoring expectations about nominal variables, while its effects on disagreement about real variables are moderate.

The extent of disagreement among forecasters may be indicative of the degree of uncertainty surrounding macroeconomic forecasts. This interpretation justifies using the dispersion of forecasts as one of the risk factors underpinning the World Economic Outlook’s fan chart for global economic growth. This is the approach taken under the new methodology for the fan chart (Kannan and Elekdag, 2009). In another area of research—predicting economic and financial crises—papers by Prati and Sbracia (2002) and Kannan and Köhler-Geib (2009) show that the dispersion of analysts’ forecasts is a significant predictor of financial crises.

There is strong evidence that consensus forecasts are inefficient and biased. Loungani (2001) showed that forecasts for both advanced and emerging and developing economies are characterized by a tendency for excessive smoothing (serial correlation in forecast revisions) and systematic biases. These results were confirmed in a more recent data set by Ager, Kappler, and Osterloh (2009). The inefficiency of forecasts is partly due to informational rigidities faced by all agents—including consumers, investors and forecasters—when forming their expectations. Coibion and Gorodnichenko (2009) show that mean forecasts fail to completely adjust on impact to structural shocks, leading to statistically and economically significant deviations from the null of full information. The behavior of forecast errors following structural shocks is consistent with the predictions of models of informational rigidities.

It is somewhat surprising that there is no evidence that the practical implications of the inefficiency of consensus forecasts are well understood by forecasters themselves. For example, forecasters fail to correct their individual forecasts for the inefficiency of consensus forecasts (Crowe, forthcoming). This finding offers an explanation for a number of empirical regularities, such as the positive short-run serial correlation observed in stock prices and the apparent success of momentum trading strategies, while posing a challenge for the efficient markets hypothesis more generally.

References

  • Ager, P., M. Kappler, and S. Osterloh, 2009, “The Accuracy and Efficiency of the Consensus Forecasts: A Further Application and Extension of the Pooled Approach,” International Journal of Forecasting, Vol. 25, No. 1, pp. 16781.

    • Search Google Scholar
    • Export Citation
  • Coibion, O., and Y. Gorodnichenko, 2009, What Can Survey Forecasts Tell Us About Information Rigidities (unpublished).

  • Crowe, Chris, forthcoming, “Forecast and Inefficient Information Aggregation,” IMF Working Paper.

  • Dovern, J., U. Fritsche, and J. Slacalek, 2009, “Disagreement among Forecasters in G7 Countries,” ECB Working Paper No. 1082 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation
  • Fintzen, D., and H. Stekler, 1999, “Why Did Forecasters Fail to Predict the 1990 Recession,” International Journal of Forecasting, Vol. 15, No. 3, pp. 30923.

    • Search Google Scholar
    • Export Citation
  • Isiklar, G., K. Lahiri, and P. Loungani, 2006, “How Quickly Do Forecasters Incorporate News? Evidence from Cross-Country Analysis,” Journal of Applied Econometrics, Vol. 21, No. 6, pp. 70325.

    • Search Google Scholar
    • Export Citation
  • Kannan, P. and S. Elekdag, 2009, “Incorporating Market Information into the Construction of the Fan Chart,” IMF Working Paper 09/178.

  • Kannan, P., and F. Köhler-Geib, 2009, “The Uncertainty Channel of Contagion,” IMF Working Paper 09/212.

  • Loungani, P., 2001, “How Accurate Are Private Sector Forecasts? Cross-country Evidence from Consensus Forecasts of Output Growth,” International Journal of Forecasting, Vol. 17, No. 3, pp. 41932.

    • Search Google Scholar
    • Export Citation
  • Loungani, P., 2002, “There Will Be Growth in the Spring: How Credible Are Forecasts of Recovery?” World Economy, Vol. 3, No. 1, pp. 16.

    • Search Google Scholar
    • Export Citation
  • Loungani, P., H. Stekler, and J. Rodriguez, 2008, “Economic Forecasts: Too Smooth by Far?” World Economy, Vol. 9, No. 2, pp. 112.

  • Loungani, P., H. Steckler, and N. Tamirisa, forthcoming, “Cross-Country Evidence on Forecasting Turning Points: Consensus and Disagreement,” IMF Working Paper.

    • Search Google Scholar
    • Export Citation
  • Prati, A., and M. Sbracia, 2002, “Currency Crises and Uncertainty About Fundamentals,” IMF Working Paper 02/3.

  • Zarnowitz, V., 1986, “The Record and Improvability of Economic Forecasting,” NBER Working Paper 2099 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
IMF Research Bulletin, March 2010
Author: International Monetary Fund. Research Dept.